
Amazon Transcribe now offers Toxicity Detection, an ML-powered feature that identifies and classifies toxic content in spoken conversations across seven categories, reducing moderation review time by 95%.


<div><p>This article announces Amazon Transcribe Toxicity Detection, an ML-powered feature that identifies toxic content in spoken conversations using audio and text analysis.</p><ul><li>Detects toxicity using both speech cues (tone, pitch) and transcribed text</li><li>Classifies toxic content across seven categories: sexual harassment, hate speech, threat, abuse, profanity, insult, graphic</li><li>Reduces moderation time by 95% by pinpointing exact toxic moments</li><li>Primarily used in online gaming and social media for player conversation monitoring</li><li>Available now for US English with batch processing</li><li>Supported in six AWS regions: US East (N. Virginia, Ohio), US West (Oregon), Asia Pacific (Sydney), Europe (Ireland, London)</li><li>Additional charges apply based on toxicity detection pricing</li></ul><p>Amazon Transcribe Toxicity Detection streamlines content moderation by automatically identifying and categorizing harmful language in audio, enabling faster action by human moderators.</p></div>


Related articles